Adverse events following acupuncture: prospective survey of 32 000 consultations with doctors and physiotherapistsBMJ 2001; 323 doi: http://dx.doi.org/10.1136/bmj.323.7311.485 (Published 01 September 2001) Cite this as: BMJ 2001;323:485
- Adrian White, senior lecturer ()a,
- Simon Hayhoe, anaesthetistb,
- Anna Hart, principal lecturerc,
- Edzard Ernst, professora
- a See Editorial by Vincent and p 486 Department of Complementary Medicine, School of Postgraduate Medicine and Health Sciences, University of Exeter EX2 4NT
- b Pain Clinic, Colchester District General Hospital, Colchester CO4 5JL
- c Faculty of Science, University of Central Lancashire, Preston PR1 2HE
- Correspondence to: A White
- Accepted 17 May 2001
Acupuncture is increasingly popular, but it is not free from risk for the patient.1 Safety is best established with prospective surveys. Our aim was to ascertain the incidence of adverse events related to acupuncture treatment, as currently practised in Britain by doctors and physiotherapists.
Participants, methods, and results
Volunteer acupuncture practitioners were recruited through journals circulated to members of the British Medical Acupuncture Society and the Acupuncture Association of Chartered Physiotherapists (approximately 2750 members).2 A prospective survey was undertaken using forms for intensive event monitoring that had been piloted previously.3 Minor adverse events were defined as “any ill-effect, no matter how small, that is unintended and non-therapeutic, even if not unexpected.” These events were reported every month, along with the total number of consultations. Minor or serious events that were considered to be “significant”—“unusual, novel, dangerous, significantly inconvenient, or requiring further information”—were reported on separate forms when they occurred. Anonymous reporting was accepted. A sample size of 30 000 consultations was necessary to identify with 95% confidence any adverse event with a frequency of 1 in 10 000 consultations.4
Estimates of incidences per 10 000 population were calculated with the acupuncturist (not the consultation) as the primary sampling unit. Since the data were skewed, with extreme values present, confidence intervals corrected for bias were calculated using bootstrapping procedure “bs” on estimates from the function “svyratio” in intercooled Stata version 6.0 with 10 000 replications.
Data were collected from June 1998 to February 2000 from 78 acupuncturists, 13 of whom chose to remain anonymous. The average age of the acupuncturists was 47 (range 27-71) years, 61% were doctors and 39% physiotherapists, and 71% had practised for five years or more. In all, 31 822 (median 318, range 5-1911) consultations were included.
Altogether, 43 “significant” events were reported (table), giving a rate of 14 per 10 000 (95% confidence interval 8/10 000 to 20/10 000). In addition, 48 apparently similar events were reported on the monthly forms, presumably due to different interpretations of “significant”. All adverse events had cleared within one week, except for one incident of pain that lasted two weeks and one of sensory symptoms that lasted several weeks. According to accepted criteria,3 none (0/10 000 to 1.2/10 000) of these events was serious.
A total of 2135 minor events was reported, giving an incidence of 671 per 10 000 (42/10 000 to 1013/10 000) consultations. The most common events were bleeding (310 (160 to 590) per 10 000 consultations) and needling pain (110 (49-247) per 10 000 consultations). Aggravation of symptoms occurred in 96 (43-178) per 10 000 consultations; in 70% of these cases, there was a subsequent improvement in the presenting complaint. The highest rates reported by individual acupuncturists, expressed as a percentage of consultations, were 53% for bleeding, 24% for pain, and 11% for aggravation of symptoms.
Doctors and physiotherapists who performed acupuncture reported no serious adverse events and 671 minor adverse events per 10 000 acupuncture consultations. These rates are classified as minimal5; however, 14 per 10 000 of these minor events were reported as significant. These event rates are per consultation, and they do not give the risk per individual patient.
Demographic data suggest that the acupuncturist volunteers were reasonably representative of the members of the two societies, but over-reporting and under-reporting are inherently possible in such studies. High individual rates may be due to a low personal threshold for reporting, or they may indicate the need for further training of the acupuncturist. Some avoidable adverse events occurred, and acupuncturists might consider modifying their practice to reduce the incidence of such events.
We thank members of the British Medical Acupuncture Society and the Acupuncture Association of Chartered Physiotherapists for collecting data, Mike Fitter and Hugh MacPherson for advice in designing the questionnaire, and Val Hopwood for help in recruiting volunteers.
Contributors: EE, SH, and AW planned the study, which was supervised by AW. The data were collected by members of the British Medical Acupuncture Society and the Acupuncture Association of Chartered Physiotherapists. The results were collated by AW, and AH performed the statistical analysis. The final report was written by AW, SH, AH, and EE. AW and EE will act as guarantors.
Funding The posts of AW and EE are funded by the Maurice Laing Foundation.
Competing interests AW has received fees for lecturing at scientific and educational meetings arranged by the British Medical Acupuncture Society and the Acupuncture Association of Chartered Physiotherapists. SH has received fees for lecturing and for acting as editor of the professional journal of the British Medical Acupuncture Society, Acupuncture in Medicine.